Computer science professor develops new methods to protect wind and solar power grids from cyber attacks

Milos Manic, Ph.D.
Milos Manic, Ph.D., IEEE Fellow, professor of computer science and director of the VCU Cybersecurity Center

Renewable power sources have potential to reduce energy costs and carbon footprint, but the massive size of wind and solar power grids presents its own set of cybersecurity challenges.

VCU Engineering’s Milos Manic, Ph.D., a fellow of the IEEE and the Commonwealth Cyber Initiative, professor of computer science and director of the VCU Cybersecurity Center, is applying his internationally recognized expertise in cybersecurity for essential systems including power grids to help make grids for wind and solar energy more secure.

Manic and colleagues from the Idaho National Laboratory (INL) are developing artificial intelligence (AI) methods to detect cyber threats in wind and solar energy grids. Their work is part of a larger project with the U.S. Department of Energy (DOE) and led by the University of Utah and INL.

This effort advances the DOE’s push for advanced cybersecurity methods to safeguard “the nation's most critical infrastructure: the power grid,” Manic said.

Wind and solar power grids must be much larger than traditional hydroelectric plants to produce the same amount of energy. This giant scale offers potential adversaries an expanded attack surface with multiple entry points.

“With grids moving toward renewables including solar and wind, the decentralized model of geographically distributed sources naturally increases potential vulnerabilities in both physical and cyber components of the grid,” Manic said.

To address this, the researchers are designing enhancements to a cyber-physical intrusion detection sensor for which they won an R&D 100 award. The original sensor used AI and machine learning techniques to improve its own effectiveness each time a would-be hacker tried to break into the system. It was widely recognized as an intelligent, integrated and highly resilient approach to recognizing and responding to threats in conventional power system environments.

The team is now modifying and dramatically scaling this sensor to make it responsive to attacks on solar grids. In the final phase, they will install the new sensor on a solar power grid’s industrial control system and subject it to penetration testing.

The researchers are also developing a data management engine to protect solar energy grids. This tool will collect and analyze data on events that present both cyber and physical anomalies and will feature a cyber-physical visualization system for metrics and ongoing threat awareness.

The team will develop algorithms to integrate anomaly detection into a variety of renewable energy systems. They will install the resulting system on hardware in a test bed and, in the final phase, demonstrate it on an operational power grid in the Pacific Northwest. They will also evaluate existing anomaly- and intrusion-detection technologies used in wind power plants and provide recommendations to the industry.

“The exciting part of our work is that we get to deploy cutting-edge AI techniques, such as physics-informed deep learning, and roll cyber and physical anomaly detection and resilience into understandable, actionable situational awareness of highly complex systems,” Manic said.